Fmriprep + tedana for multi-echo data leads to massive signal drop out

Software Support
,
1

Summary of what happened: I ran fMRIprep on multi-echo data preprocessed with fmriprep using the flag –me-output-echos and then processed the output echos (in native space) with tedana. The tedana output (every image) has an odd, stripey drop-out and I can’t seem to understand why. Screenshot is of the adaptive mask, and all post-tedana subject images have the same drop-out pattern. Any ideas of why this might be happening? Thanks in advance for your guidance!!

Command used (and if a helper script was used, a link to the helper script or the command generated):

>tedana -d ./fmri_preproc/derivatives/fmriprep_out/sub-pilot2decob/func/sub-pilot2decob_task-decobi_run-1_
echo-1_desc-preproc_bold.nii.gz ./fmri_preproc/derivatives/fmriprep_out/sub-pilot2decob/func/sub-pilot2de
cob_task-decobi_run-1_echo-2_desc-preproc_bold.nii.gz ./fmri_preproc/derivatives/fmriprep_out/sub-pilot2d
ecob/func/sub-pilot2decob_task-decobi_run-1_echo-3_desc-preproc_bold.nii.gz ./fmri_preproc/derivatives/fm
riprep_out/sub-pilot2decob/func/sub-pilot2decob_task-decobi_run-1_echo-4_desc-preproc_bold.nii.gz ./fmri_preproc/derivatives/fmriprep_out/sub-pilot2decob/func/sub-pilot2decob_task-decobi_run-1_echo-5_desc-preproc_bold.nii.gz -e 12.200 28.920 45.64
0 62.360 79.080 --prefix sub-pilot2decob_task-decobi_run-1_desc-tedana --mask ./fmri_preproc/derivatives/
fmriprep_out/sub-pilot2decob/func/sub-pilot2decob_task-decobi_run-1_desc-brain_mask.nii.gz --masktype none --tedpca kundu --out-dir ./fmri_preproc/derivatives/tedana/sub-pilot2decob/func/sub-pilot2decob_task-decobi_run-1 --verbose

Version:

tedana 25.0.1

Environment (Docker, Singularity / Apptainer, custom installation):

Singularity

Data formatted according to a validatable standard? Please provide the output of the validator:

BIDS

Screenshots / relevant information:

tedana-adaptive-mask
tedana-adaptive-mask1236×1042 65.6 KB

2

I’m noticing that your script includes --masktype none That means the adaptive mask is just removing NaN values and values <=0. (FWIW, the default setting is --masktype dropout which is what I’d generally recommend). That said, given adaptive masking shouldn’t be doing much, it will only depend on your mask: ./fmri_preproc/derivatives/ fmriprep_out/sub-pilot2decob/func/sub-pilot2decob_task-decobi_run-1_desc-brain_mask.nii.gz and your time series ./fmri_preproc/derivatives/fmriprep_out/sub-pilot2decob/func/sub-pilot2decob_task-decobi_run-1_echo-?_desc-preproc_bold.nii.gz Can you check all 5 echo volumes to see if those data look ok or if they have negative or Nan values that match what you’re seeing in the above output? Make sure to check across because it’s possible a single corrupted volume might be causing downstream problems.

Hope this helps.

Dan

3

Hi @handwerkerd, thank you very much for your reply! Indeed, we had some noise volumes causing issues downstream. I sat down today with @CesarCaballeroGaudes and we fixed the issue. I greatly appreciate the time you took to help me out.

1 Like